a thoughtful web.
Good ideas and conversation. No ads, no tracking.   Login or Take a Tour!
comment by amonico48
amonico48  ·  3192 days ago  ·  link  ·    ·  parent  ·  post: Efficiency in contemporary software...

What are your core competencies when it comes to computer programming or software engineering? You've got to be an assembly language or a c programmer to readily perceive the analysis. If you wrote computer programs before 1999 this fact wouldn't be lost on you. The same is true for game developers or other performance critical devepors today. Take time to read the article or blog publication carefully and do some research; talk to some more people.

What do you understand by processor speed of 3.5 gigahertz? Some years ago the best you could get was 66megahertz.





CrazyEyeJoe  ·  3192 days ago  ·  link  ·  

As a matter of fact, writing C and assembler is what I've been doing the past two years. Indeed I'd just graduated before then, so my experience is still limited. However, I've optimised plenty of code.

Going from C to assembler, I seem to remember cutting down processing time by half, or something around that range. Maybe a bit more in some cases. That's one or two orders of magnitude below what you suggest.

Honestly though, my criticism isn't so much with the numbers as it is with the content of the article. I felt like you spent a lot of time taking about trivialities and hearsay, and not much time developing your argument with facts.

Yes, speed (obviously) matters, power consumption (obviously) matters, and latency (obviously) matters. No one in their right mind would say otherwise. However, so does development time, portability and maintainability. Not all software needs to be written in assembler.

Maybe I'm not the target audience for this article. Still, I feel like you wrote a lot of words to say very little. I'll admit I was rather hostile in my response, and I'm sorry that I was so aggressive. I was simply disappointed with the article.

amonico48  ·  3192 days ago  ·  link  ·  

How do you calculate latency in seconds given aggregated overhead and a given processor speed? In my post I explained how to calculate the overhead in a loop that is supposed to run a million times.